This PDF file contains the front matter associated with SPIE Proceedings Volume 6501, including the Title Page, Copyright information, Table of Contents, and Conference Committee listing.

A 111-Mega pixel, 92×92 mm², full-frame CCD imager with 9×9 um² pixel size has been developed for use in scientific applications. Recent interest for ultra-high resolution imagers for electronic imaging OEM customers in various scientific markets, including biotechnology, microscopy, crystallography, astronomy, spectroscopy, and digital photography markets has lead to the development of the STA1600A 111-Mega pixel monochromatic charge-coupled device. Innovative design techniques were utilized in the early development of this device, yielding low RMS noise and high MTF for readout speeds ranging from 1 Mpixel/s to 10 Mpixel/sec. This paper will provide detailed information on the design and performance capabilities of the STA1600A, as well as background information on the commercial uses of this device.

Dark current is caused by electrons that are thermally exited into the conduction band. These electrons are collected by the well of the CCD and add a false signal to the chip. We will present an algorithm that automatically corrects for dark current. It uses a calibration protocol to characterize the image sensor for different temperatures. For a given exposure time, the dark current of every pixel is characteristic of a specific temperature. The dark current of every pixel can therefore be used as an indicator of the temperature. Hot pixels have the highest signal-to-noise ratio and are the best temperature sensors. We use the dark current of a several hundred hot pixels to sense the chip temperature and predict the dark current of all pixels on the chip. Dark current computation is not a new concept, but our approach is unique. Some advantages of our method include applicability for poorly temperature-controlled camera systems and the possibility of ex post facto dark current correction.

A pixel structure suitable for an interline transfer charge-coupled device (IT-CCD) image sensor with a pixel size of 2 &mgr;m square or less was developed. The first feature relates to the structure of a transfer electrode, which consists of island transfer electrodes and horizontally continuous transfer electrodes, alternately placed in the form of a single layer. The island electrode is connected with polysilicon wiring formed on the horizontally continuous electrode. This pixel structure eliminates one of two electrodes on the channel stop region of the conventional single layer electrode structure. Over 30% Expansion of the photosensitive areas of 2 &mgr;m square pixel compared to the conventional design was achieved. Carrier crosstalk between adjacent pixels, often caused by application of read-out voltage to the channel stop region, is successfully suppressed in the new structure. The second feature comes from the photo-shield shape around the photodiode. By removing the extension of the photo-shield and reducing the thickness of the lateral side of the photo-shield, the fill factor can be expanded by 70%. In addition, the eclipse of incident light by the photo-shield above the vertical register is suppressed. These developments have led to the achievement of high performance in photo-sensitivity, especially in the small F number region, sensitivity shading, handling charge of the photo diode, and have provided a beneficial solution for sub-2 &mgr;m square pixel IT-CCD.

As a consequence of the reduction in the pixel sizes of charge coupled device (CCD) image sensors, the sensitivity of these sensors has decreased, which means that their signal-to-noise ratio (SNR) has also decreased even though the amount of noise is kept constant. In order to maintain and even increase the SNR, we evaluated a simulation method for estimating the sensitivity and smear noise. Smear noise and sensitivity are defined by the number of electrons in vertical registers and photodiodes, respectively. We used a finite-difference time-domain (FDTD) method to simulate the light energy which is proportional to electrons generated in a Si substrate. Using this simulation, we were able to estimate sensitivity and smear noise accurately and optimize the structure of on-chip lenses (OCLs) with respect to these parameters. When we optimized an OCL structure for an interline transfer (IT)-CCD having 1.86-&mgr;m-square pixels, we found that the optimal thickness of the OCL in regards to the smear noise was 0.25 &mgr;m thinner than the optimal thickness for the sensitivity. This result demonstrates that when designing the structures of image sensors, including the OCL shape, it is not only necessary to consider the sensitivity, but it is also important to take the smear noise into consideration.

Due to aggressive scientific specifications, Semiconductor Technology Associates and the University of Arizona's Imaging Technology Laboratory have collaborated to develop a fully depleted back illuminated CCD for scientific imaging. These devices are designed to target increased quantum efficiency into the near-infrared, without reduction in the modulation transfer function, charge transfer efficiency, or rms noise. The STA1700 series imagers are back illuminated 100 micron thick devices with a 10 micron pixel pitch targeted to meet the requirements of the Large Synoptic Survey Telescope (LSST). Recent characterization results will be presented including the point spread function of a 2 micron spot. Also discussed will be the thinning and packaging developments for the STA1700. These efforts include the addition of a backside bias contact, invar package design with high density connectors, as well as etching and backside coating optimization for high resistivity silicon.

Recent development efforts on the orthogonal transfer array (OTA) for the Pan-STARRS gigapixel camera 1 (GPC1) are described. A redesign of the prototype OTAs has been completed, and fabrication and packaging of the devices for the GPC1 are nearly complete. We briefly review the final design features and the resolution of the performance issues that arose in the first prototype devices. We then describe the packaging of the device and the challenges arising in achieving the necessary flatness at the device operating temperature. Plans and schedule for deploying focal-plane arrays of these devices are described.

We enhanced the photoelectric conversion efficiency of red light in a 15-&mgr;m-thick HARP film without deteriorating image pick-up characteristics or reliability. To achieve a higher photoelectric conversion efficiency for red light, we designed a new film structure with an increased amount of doped Te, which has a narrower band gap than that of a-Se. The thickness of the LiF-doped layer for trapping holes was increased from that of the conventional red-extended HARP film in order to weaken the internal field that would otherwise be enhanced by trapped electrons in extra doped Te. The new red-extended HARP film achieved a photoelectric conversion efficiency for red light of about 22.5% at a wavelength of 620 nm, which is twice that of the conventional red-extended film. We confirmed an improvement in signal to shot noise ratio of 3 dB and a dramatic improvement in color reproduction when we experimented with an HDTV camera with a camera tube incorporating the new film.

Miniaturized camera systems are an integral part of today's mobile phones which recently possess auto focus functionality. Commercially available solutions without moving parts have been developed using the electrowetting technology. Here, the contact angle of a drop of a conductive or polar liquid placed on an insulating substrate can be influenced by an electric field. Besides the compensation of the axial image shift due to different object distances, mobile phones with zoom functionality are desired as a next evolutionary step. In classical mechanically compensated zoom lenses two independently driven actuators combined with precision guides are needed leading to a delicate, space consuming and expansive opto-mechanical setup. Liquid lens technology based on the electrowetting effect gives the opportunity to built adaptive lenses without moving parts thus simplifying the mechanical setup. However, with the recent commercially available liquid lens products a completely motionless and continuously adaptive zoom system with market relevant optical performance is not feasible. This is due to the limited change in optical power the liquid lenses can provide and the dispersion of the used materials. As an intermediate step towards a continuously adjustable and motionless zoom lens we propose a bifocal system sufficient for toggling between two effective focal lengths without any moving parts. The system has its mechanical counterpart in a bifocal zoom lens where only one lens group has to be moved. In a liquid lens bifocal zoom two groups of adaptable liquid lenses are required for adjusting the effective focal length and keeping the image location constant. In order to overcome the difficulties in achromatizing the lens we propose a sequential image acquisition algorithm. Here, the full color image is obtained from a sequence of monochrome images (red, green, blue) leading to a simplified optical setup.

In a longstanding effort to overcome limits of film and the charge coupled device (CCD) systems in electron microscopy, we have developed a radiation-tolerant system that can withstand direct electron bombardment. A prototype Direct Detection Device (DDD) detector based on an Active Pixel Sensor (APS) has delivered unprecedented performance with an excellent signal-to-noise ratio (approximately 5/1 for a single incident electron in the range of 200-400 keV) and a very high spatial resolution. This intermediate size prototype features a 512×550 pixel format of 5&mgr;m pitch. The detector response to uniform beam illumination and to single electron hits is reported. Radiation tolerance with high-energy electron exposure is also impressive, especially with cooling to -15 °C. Stable performance has been demonstrated, even after a total dose of 3.3×10⁶ electrons/pixel. The characteristics of this new detector have exciting implications for transmission electron microscopy, especially for cryo-EM as applied to biological macromolecules.

I have developed for cameras a new system of capturing images called the "optical-multiplex" system. The system is comprised of an image sensor and a multi-lens array (or an array of pin holes). This system has the advantages of being compact and light and being able to provide a deep depth of field. In a model of the system for simulation purposes, light passes through five pin holes in an "aperture sheet" and the resulting object information is detected by the image sensor and processed by the signal-processing unit, which outputs the "optical-multiplex" signal. The simulation model incorporating both signal and noise shows that most pixels in this new system have a better signal-to-noise ratio than in the conventional single-lens system.

Inspired by the natural phenomenon of hyperacuity, a novel approach has been analyzed that allows to access highly accurate information with an artificial apposition compound eye despite the number of image pixels is small. This is achieved by oversampling of the object space due to overlapping fields-of-view of adjacent optical channels. The first approach uses the knowledge about the impulse response function derived by linear system theory to calculate the position of objects like point sources and edges from the measured optical powers in adjacent channels. Therefore, the implementation of a precise position detection with an accuracy increase of up to 50 times compared to the conventional image resolution is demonstrated. The second approach that is presented, works in a more general way because it is independent of the exact imaging model. With the help of the latter, the accuracy of the position detection of an edge was increased by a reproducible factor of 25. As presented here, the second approach also enables the measurement of a width with sub-pixel accuracy.

We describe the design and implementation of a high dynamic range (HDR) imaging system capable of capturing RGB color images with a dynamic range of 10,000,000 : 1 at 25 frames per second. We use a highly programmable camera unit with high throughput A/D conversion, data processing and data output. HDR acquisition is performed by multiple exposures in a continuous rolling shutter progression over the sensor. All the different exposures for one particular row of pixels are acquired head to tail within the frame time, which means that the time disparity between exposures is minimal, the entire frame time can be used for light integration and the longest exposure is almost the entire frame time. The system is highly configurable, and trade-offs are possible between dynamic range, precision, number of exposures, image resolution and frame rate.

It is well known as the presence of fine and ultra-fine particulate solids products (fillers), inside a concrete, strongly contributes to change final manufactured products characteristics. Fine carbonate fillers, in fact, complement the deficiency in fine particles of the cement's particle size distribution, which can enhance both the flowability and stability of fresh concrete. They also fill in between the relatively coarser cement grains, reducing the room available for water and consequently the water demand. In conventional concrete mixtures, slight reductions in the setting time have been often reported when carbonate fillers were used, without significant effects on the workability. However, as the use of high-performance concrete continues to rise, carbonate fillers are being added in low water/cement ratio superplasticized-mixtures. In this paper an innovative approach, based on thermal-imaging, was applied in order to establish a correlation, for the different utilized fillers, among fillers quantity and quality, concrete behaviour during ageing and final mechanical characteristics of the products at 28 days.

We have developed a pixel unit for CMOS image sensors (CISs) that has a shared transistor architecture with diagonally connected pixels. This pixel unit is composed of four photodiodes and seven transistors. It has a pixel size of 2.5-&mgr;m square. The transistors were designed using 0.18-micron aluminum process technology. Shared diffusion for reading signal electrons occurs between the corners of two photodiodes. The advantages of this layout include a long amplifier gate length and a large photodiode area.

We have been investigating a system to detect moving objects correctly at the place where luminous intensity is changing because of the influence of incident light such as sunlight, fluorescent light and car headlight. The object detection system consists of a smart image sensor and a modulated LED light, and it is possible to suppress the influence of the change of background light by using a different value between two image values when the LED light is turned on and off. Because the speed of modulation is high for accurate detection of moving objects, electric charges from a photodiode are distributed into two capacitors by switching in sync with the LED light in a pixel circuit of the sensor. Also, the sensor has a subtraction function by a current mirror circuit to reduce the same charges from two capacitors before saturation. By the frequent subtractions, it is possible to increase only the influence of the modulated light and reconstruct wide dynamic range images at outside of the sensor by using the information of subtractions and the voltage value of each capacitor.

This paper proposes and demonstrates a linear-logarithmic image sensor that can operate in both the linear and logarithmic modes, and exhibits low noise and no flickers. Its pixel is composed of one buried-photodiode, five n-channel MOS transistors, two p-channel MOS transistors, and one capacitor. During the linear mode operation, the pixel behaves the same as the four-transistor one, offering a low dark current and low noise. Also, in the logarithmic mode, it operates in an integrating manner in contrast to the majority of logarithmic-response CMOS image sensors that operate in a continuous manner. Thus it has no flickers and can calibrate the fixed pattern noise using the reference level. A wide dynamic range of 190 dB has been confirmed with a 256 × 256-pixel image sensor employing this novel pixel architecture.

Our new type of camera module is small and thin, and is able to detect the distances of objects as well as their images. The module comprises a four-lens array, one imaging sensor and optical filters. The imaging sensor is divided into four areas, two of which are covered with green filters and the other two with infiared filters. A prototype was fabricated with a focal length of 2.63 mm and a baseline length of 2.59 mm. The two images with the same optical filters have parallax, so the distances of objects can be calculated by comparing the two images. We use infrared images illuminated with infrared LEDs at night, and green images during the daytime. After calibrating the images, we achieved a distance-detection accuracy of within ±2.5% at 1 m in spite of the camera's small baseline length. Consequently, our new distance detection camera module is small and thin, and generates a depth-image of an object as well as its image. Our camera module is thus applicable to vehicles, security systems and three-dimensional imaging.

A novel principle has been developed to build an ultra wide dynamic range digital CMOS image sensor. Multiple integrations are used to achieve the required dynamic. Its innovative readout system allows a direct capture of the final image from the different exposure time with no need of external reconstruction. The sensor readout system is entirely digital, implementing an in-pixel ADC. Realized in the STMicroelectronics 0.13&mgr;m CMOS standard technology, the 10&mgr;m x 10&mgr;m pixels contain 42 transistors with a fill factor of 25%. The sensor is able to capture more than 120dB dynamic range scenes at video rate.

To realize a low-voltage CMOS imager with a small pixel size, we have proposed a new pixel structure composed of only three transistors without any circuit sharing technique. The pixel has a gate-common transistor that compares a photodiode voltage on the gate node with a ramp signal on the source node to perform a single-slope A/D conversion based on a pulse-width-modulation pixel-reading scheme. The large gain of the in-pixel comparator contribute to the small input-referred noise and surpress column-to-column fixed-pattern-noise (FPN). Pixel-to-pixel FPN is suppressed by a feedback reset. Our CMOS imager can lower the operating voltage with less degradation of the dynamic range than that of ordinary active pixel sensors. We have fabricated a 128×96-pixel prototype sensor with an on-chip ramp generator and bootstrap circuits in a 0.35-&mgr;m CMOS technology, and successfully demonstrated its operations with a 1.5-V single power-supply voltage.

A CMOS image sensor with highly accurate object extraction pre-processing functions by 960-fps sub-sampling operation, 12-bit column parallel successive approximation ADCs and column parallel ALUs has been developed. The pixel is composed of four transistors type pixel which shares the source follower transistor and the pixel select transistor. The each ADC is composed of the noise and signal holding capacitance, the noise reduction circuit, the comparator and the small DAC that combined both the reference voltage ratios and capacitance ratios. In the ALU, the object categorization pre-processing is performed by the each macro block of 3 × 3 pixels which has a reference pixel and its neighboring eight pixels. The three image features which are the edge of object, the direction of edge-vector and the average of light-intensity of 3 × 3 pixels corresponded to each pixel are extracted by the ALUs. The image and the results of the object extraction pre-processing are outputted by every 60-fps. The image sensor was fabricated by 0.35-&mgr;m 2P3M technology. The pixel pitch is 5.3-&mgr;m, the number of pixels is 640H × 360V and the chip size is 4.9-mm square.

A temperature resistant wide dynamic range (WDR) CMOS image sensor has been developed using the very low dark current front-end of line (VLDC FEOL) and the metal hermetic seal ceramic leadless chip carrier (CLCC) package suppressing the degradation of the spectra response of the on-chip micro lens and color filter (OCML/OCCF). A 1/4 inch VGA 5.6 &mgr;m pixel pitch WDR CMOS image sensor has been fabricated through 0.18 &mgr;m 2P3M process with the VLDC FEOL which contains the pinned photodiode with less electrical fields, the less plasma etching damages, the transfer gate with the suppressed current at Si-SiO₂ interface and the furnace temperature process for the re-crystallization. The sensor chips with the conventional OCML/OCCF assembled into the metal hermetic seal package by the low residual oxygen vacuum welding machine has finally received the thermal stress test (150 deg.C/500 hours). The dark current is 350 pA/cm² at 85 deg.C (50 pA/cm² at 60 deg.C). No degradation of the spectra response in any of R/G/B pixels is observed after the thermal stress test. It is found that the thermal decomposition of the OCML/OCCF (phenol resin) is not caused easily in nitrogen with the low residual oxygen concentration. The sample images captured by the WDR CMOS image sensor assembled camera keep good quality up to 85 deg.C.

We present characterization results for a 5 million pixel CMOS image sensor designed for high speed applications. This sensor is capable of outputting 14 frames per second and incorporates on-chip 12-bit digitization. We present measurements of system gain, read noise, dark current, charge capacity, linearity, photo response non-uniformity, defects, and quantum efficiency. The image sensor incorporates exposure control functionality, windowing, on-chip binning, anti-blooming capability and rolling shutter architecture to implement image capture mode. The results show a favorable aspect of the ability to achieve high speed, high resolution, and very good sensitivity in a monolithic CMOS sensor. Architecture trades for high speed imaging systems utilizing CCDs and CMOS sensors are also presented.

Polarized light is a naturally occurring phenomenon that human vision does not discern, yet it can provide useful supplementary information from an image or optical system. Polarization detection can be implemented using hybrid sensors where additional polarizing material is mounted onto a standard sensor. However these types of sensor tend to be expensive, requiring extra manufacturing and materials. Presented is a low cost polarization sensor which is implemented using standard CMOS technology and manufacturing techniques, without the need for supplementary implants or optical layers. The polarization sensor is realised using a polarization grating, formed from a standard metal layer, above a CMOS sensor. To compensate for the loss of photons due to the polarization grating, a high dynamic range sensor is implemented using large, 110 micron photodiodes. The photosensor is used in a "light to frequency conversion pixel" where the photocurrent is converted to a digital square wave output with a frequency proportional to the photon flux density. A modulation depth of 10% is achieved. A rotary encoder application implementing the polarization sensor is discussed.

On-die optics have been proposed for imaging, spectral analysis, and communications applications. These systems typically require extra process steps to fabricate on-die optics. Fabrication of diffractive optics using the metal layers in commercial CMOS processes circumvents this requirement, but produces optical elements with poor imaging behavior. This paper discusses the application of Wiener filtering to reconstruction of images suffering from blurring and chromatic aberration, and to identification of the position and wavelength of point sources. Adaptation of this approach to analog and digital FIR implementations are discussed, and the design of a multispectral imaging sensor using analog FIR filtering is presented. Simulations indicate that off-die post-processing can determine point source wavelength to within 5% and position to within ±0.05 radian, and resolve features 0.4 radian in size in images illuminated by white light. The analog hardware implementation is simulated to resolve features 0.4 radian in size illuminated by monochromatic light, and 0.7 radian with white light.

We have proposed a new type of camera module with a thin structure and distance-detection capability. This camera module has a four-lens-array with diffraction gratings (one for blue, one for red, and two for green). The diffraction gratings on the mold are formed mechanically, and the plastic lens array is fabricated by injection molding. The two green images are compared to detect parallax, and parallax-corrected blue, red and green images are then composed to generate a color image. We have developed new design software and molding technologies for the grating lenses. The depth and period of blazed gratings and the shapes of aspheric lenses are optimized; and blue, red and two green aspheric lenses with gratings are molded as a single four-lens-array. The diffraction gratings on both surfaces of each lens act to improve field curvature and realize wide-angle imaging. However, blazed gratings sometimes cause unnecessary diffraction lights that impede the formation ofhigh-resolution images. We have developed a new method to measure necessary first-order diffraction lights and unnecessary diffraction lights separately. Use of this method allows the relationship between molding conditions and necessary/unnecessary diffraction lights to be shown. Unnecessary diffraction lights can be diminished by employing the optimal molding processes, allowing our grating lenses to be used for image capture.

In the present work, the temperature versus thermal deformation (strain) with respect to time, of different coating films were studied by a non-destructive technique (NDT) known as shearography. An organic coating, i.e., epoxy, a white enamel, and a yellow Acrylic Lacquer on a metallic alloy, i.e., carbon steels, were investigated at a temperature range simulating the severe weather temperatures in Kuwait especially between the daylight and the night time temperatures, 20-60 °C. The investigation focused on determining the in-plane displacement of the coatings, which amounts to the thermal deformation (strain) with respect to the applied temperature range. Furthermore, the investigation focused on determining the thermal expansion coefficients of coatings, the slope of the plot of the thermal deformation (strain) versus the applied temperature range. In other words, one could determine, from the decreasing value of the thermal expansion coefficients of coatings, a critical (steady state) value of the thermal expansion coefficients of coatings, in which the integrity of the coatings can be assessed with respect to time. In fact, determination of critical (steady state) value of the thermal expansion coefficients of coatings could be accomplished independent of parameters, i.e., UV exposure, Humidity, exposure to chemical species, and so on, normally are considered in conventional methods of the assessment of the integrity of coatings. In other words, with the technique of shearography, one would need only to determine the critical (steady state) value of the thermal expansion coefficients of coatings, regardless of the history of the coating, in order to assess the integrity of coatings. Furthermore, results of shearography indicate that the technique is very useful NDT method not only for determining the critical value of the thermal expansion coefficients of different coatings, but also the technique can be used as a 2D- microscope for monitoring the deformation of the coatings in realtime at a submicroscopic scale.

Previously, we showed that two- and three-band color-mixing techniques could be used to achieve results optically equivalent to two- and three-band ratios that are normally implemented using multispectral imaging systems, for enhancing identification of single target types against a background and for separation of multiple targets by color or contrast. In this paper, a prototype of a wavelength-changeable two- and three-band color-mixing device is presented and its application is demonstrated. The wavelength-changeable device uses changeable central wavelength bandpass filters and various filter arrangements. The experiments showed that a color-mixing technique implemented in a pair of binoculars coupled with an imager could greatly enhance target identification of color-blindness test cards with hidden numbers and figures as the targets. Target identification of color blindness cards was greatly improved by using twoband color-mixing with filters at 620 nm and 650 nm, which were selected based on the criterion of uniform background. Target identification of a different set of color blindness test cards was also improved using three-band color-mixing with filters at 450 nm, 520 nm, and 632 nm, which were selected based on the criterion of maximum chromaticness difference. These experiments show that color-mixing techniques can significantly enhance electronic imaging and visual inspection.

Using a simple non-destructive and non-interferometric technique, we obtained an approximation of the average index profile for a partially symmetrical air-silica microstructured optical fiber. The method proposed in [1] for conventional fibers, where an image of the phase gradients is introduced into a transmitted optical field by a fiber sample, was used. An image of the phase gradients was obtained using a technique based on bright field microscopy. Then, an average of the refractive index profile for optical fibers was reconstructed using the inverse Abel transform.

A low-cost sensor platform (MORES Sensor) was combined with a microcontroller to build up an embedded solution which e.g. allows for a small hand-held Color Estimation System for blind people. The color sensor used here measures the intensity response of a surface caused by radiation with a specific wavelength in the range of visible light. This radiation is realized by means of three LEDs, red, green, and blue, so that the response intensity values create an R'G'B' color space, which differs from the standardized RGB color space due to the wavelengths of the LEDs. By adjusting the measured response of the LEDs to the known spectral response of the individual color panels of the Macbeth Color Checker Chart (MCCC) a corresponding set of coordinates can be constructed for this particular R'G'B' color space. Owing to this approach, it is possible to obtain reasonable color classification results, which can be compared to those of far more complex and expensive systems. The verification of the results was done by using the standardized MCCC together with commercial vision solutions (RGB camera in combination with PC software). Moreover, some comparison tests also prove the practicality of the here described low-cost color sensor solution.

In this paper, a vision chip for a contrast-enhanced image based on a structure of a biological retina is introduced. The key advantage of this structure is high speed of signal processing. In a conventional active pixel sensor (APS), the charge accumulation time limits its operation speed. In order to enhance the speed, a logarithmic APS was applied to the vision chip. By applying a MOS-type photodetector to the logarithmic APS, we could achieve sufficient output swing for the vision chip in natural illumination condition. In addition, a CMOS buffer circuit, a common drain amplifier, is commonly used for both raw and smoothed images by using additional switches. By using the switch-selective resistive network, the total number of MOSFETs for a unit pixel and the fixed-pattern noise were reduced. A vision chip with a 160×120 pixel array was fabricated using a 0.35 &mgr;m double-poly four-metal CMOS technology, and its operation was experimentally investigated.

In this paper, a new CMOS image sensor is presented, which uses a PMOSFET-type photodetector with a transfer gate that has a high and variable sensitivity. The proposed CMOS image sensor has been fabricated using a 0.35 &mgr;m 2-poly 4- metal standard CMOS technology and is composed of a 256 × 256 array of 7.05 × 7.10 &mgr;m pixels. The unit pixel has a configuration of a pseudo 3-transistor active pixel sensor (APS) with the PMOSFET-type photodetector with a transfer gate, which has a function of conventional 4-transistor APS. The generated photocurrent is controlled by the transfer gate of the PMOSFET-type photodetector. The maximum responsivity of the photodetector is larger than 1.0 × 10³ A/W without any optical lens. Fabricated 256 × 256 CMOS image sensor exhibits a good response to low-level illumination as low as 5 lux.

An X-ray direct-conversion type detector with a spatial resolution of 10-11 &mgr;m was developed for real-time biomedical imaging. The detector incorporates the X-ray SATICON pickup tube with a photoconductive target layer of amorphous selenium. For high-resolution imaging, the X-ray image is directly converted into an electric signal in the photoconductive layer without image blur. Microangiography experiments were carried out for depicting angiogenic vessels in a rabbit model of cancer using the direct-conversion detector and a third generation synchrotron radiation source at SPring-8. In synchrotron radiation radiography, a long source-to-object distance and a small source spot can produce high-resolution images. After transplantation of cancer cells into the rabbit auricle, small tumor blood vessels with diameters of 20-30 &mgr;m in an immature vascular network produced by angiogenesis were visualized by contrast material injection into the auricular artery at a monochromatic X-ray energy of 33.2 keV just above the iodine K-edge energy. The synchrotron radiation system is a useful tool to evaluate the micro-angioarchitecture of malignant tumors in animal models of cancer for in vivo preclinical studies.

Detection of moving or novel objects appearing on transportation units is one of standard problems for industrial vision applications. Especially for real-time embedded applications in industrial environment constraints like low-cost and low-power require a proper trade-off between this critical design issues and robustness of the device. This paper presents an embedded system implementation of an algorithm for realtime extraction of moving objects based on focal-plane asynchronous processing of relative changes in pixel intensity using address-event representation (AER) protocol, and its implementation in form of low-cost, small size and low-power embedded system.